G-Diffuser: Exploring creative possibilities with Stable Diffusion in/out-painting

Name of Project: G-Diffuser

Proposal in one sentence: G-Diffuser is a scriptable / extensible interface for diffusion models through the open Stable Diffusion gRPC Server project.

Description of the project: G-Diffuser has several components such as an extensible interactive CLI and Discord bot, but the focus of the project is on exploring the possibilities of in/out-painting using the state-of-the-art implementations in the sdgrpcserver project. For example the “zoom maker” CLI script combines out-painting with high quality compositing to automatically generate smooth infinite zoom animations given only a single text prompt:

Grant Deliverables: New types of animations with more camera movement and perhaps a “perspective-correct” zoom that looks less like an infinite zoom and more like a camera fly-through.

Squad Lead: Christopher Friesen
Twitter: @parlance_zz
Discord: parlance#1220


Sorry, I have to add this in a reply because new accounts can only submit 2 links per post apparently.

Github page for the project is here: GitHub - parlance-zz/g-diffuser-bot: Discord bot and Interface for Stable Diffusion

This project resembles @metamyth 's Lucidly around these forums. If I’m not wrong its perspective widening too. You should definitely connect with him.

also the outpainting is cool. maybe you can allow people to freeze and comment at moments. this would be similar to the soundcloud ux, letting people co-like and form bonds because it’s all an art here at the end. community play also comes active with a lot of web3 art folks and project’s here :smiley: